MakeoverMonday: MakeoverMonday!

Cohort Analysis
Download the v10 workbook here.

We’re at halfway. Over 200 of you have posted more than 1,000 makeovers over 25 weeks. This is simply amazing.

Where else could you find a growing source of charts and data to play with?

Where else could you find so many different ways of telling data stories?

When Andy and I started this, we thought it’d just be us goofing around. But it’s you, the community, who have made this something more special than we could have imagined.

Thank you!

This week we’re making over MakeoverMonday data. I wanted to do some cohort analysis (check out a great post on using LOD calcs for this here). It turns out that those of you who’ve done the first 5 weeks of Makeovers contribute about 50% of all makeovers each week. Below is a percent of total view of the chart above:

Cohort Analysis (%)

Keep it up, gang! I’m loving seeing all the incredible ideas you come up with each week.

quick snap



MakeoverMonday: Am I safe in Japan?

I’m very excited to be in Tokyo this week. I’ll be presenting the Tableau 10.0 roadshow and at a bunch of partner and customer events, too. I’m very thankful for this opportunity!

The makeover

This week, Andy K found a chart showing reported thefts in Japan. The chart we’re focusing on shows the number of reported thefts in Japan in 2012. I thought I’d make it over to make it personal. Personal to me. Could I use this to prove the reputation Japan has of being a safe country to visit?

Will I be safe in Japan dash
Download the workbook here. (Tableau v10)

I didn’t start off wanting to ask that question. My original starting point was to draw a straightforward treemap. I haven’t seen too many of these in MakeoverMonday so I thought they deserved some attention.

Treemap: only ok
Treemap: only ok

The treemap was ok, but it didn’t amaze me, and I didn’t feel that I’d really hit on anything interesting for MakeoverMonday. All I’d done was take sectors of a circle and make them sub-rectangles of a bigger one.

As I interacted with the data, though, I realized I could look for patterns relevant to me. This week is my first visit to Japan. As the original article describes, Japan has a reputation for being safe. “Well then,” I thought, “which of these crimes could I fall prey to and are they common?”

That led to my makeover and a personal story to prove the relative safety of Japan, using available data. Given there are so few crimes, it’s fair to say that this data supports the reputation Japan has of being a safe place to visit.

Iterations and alternatives

Do I love this treemap? Not especially: in fact, with this dataset, I think a pie makes it easier to see the proportional to whole relationship than a treemap. Look at the figure below. In which chart is it easier to see that Vehicle Theft accounts for about 30% of reported theft. [Note: the previous statement comes with the normal caveats about pies and their problems. I know the problems with pies, you don’t have to tell me them; I’m just describing my thought process as I explored the data and built different views.]

Pie or Treemap?
Pie or Treemap?

If I was going for efficiency in the makeover, I’d probably have chosen a stacked bar or even a normal bar chart. These allow for the easiest lookup of data. Here they are below.

stacked bar bar

The original chart

Here’s the original chart and my thoughts on it:

What I liked

  1. Everything is labelled, so I can lookup any value I want
  2. There’s a total in the middle so I can the proportions and relate it to the entire number of reported thefts
  3. The labels are aligned making them easier to lookup than otherwise

What I didn’t like

  1. It’s a sunburst chart. There’s a certain pleasure in looking around and following shapes from the centre outwards, but it’s so slow and inefficient. A normal bar chart gets the job done quicker
  2. The inner label shows the actual total number of reported thefts, but the outer numbers show percentages. That’s not made clear.
  3. The outer level of the sunburst appears to be randomly sorted. It could have been in descending or alphabetical order.


MakeoverMonday: Women in the workplace

This week MakeoverMonday is LIVE at Tableau Conference on Tour. Check out the hastags #makeovermonday and #data16 during Monday to follow things live.

Women in the workplace

For my makeover this week, I wanted to simplift the message. The differences between 2012 and 2015 weren’t that great. There are more women at each level, but the trends themselves haven’t changed. I decided to remove 2012 from my data to focus more clearly on the Pipeline story.

I liked the quote in the first paragraph of the original so lifted that for the title.

In our makeover about women in legislature, I extended the y-axis to 100% to emphasise the distance to parity with men. In this case, I decided to end the y-axis at 50%. To make it clear that the top of the chart is 50% I made the reference line stand out, and put the title beneath it. Did that succeed? Did you see the reference line?

The original

Well, it’s a pipeline. Of sorts.

The original chart wasn’t a great one this week.

What I liked:

  • There’s a table, so I can lookup the numbers
  • The colour scheme is very easy to distinguish
  • They attempted to use a visual metaphor for a pipe

What could have been improved:

  • The mix of line chart and pipeline renders the chart pretty meaningless: it’s not possible to see what’s actually being shown in the chart
  • The designers appear to have drawn a straight line in the chart, but the data doesn’t quite drop the way it’s shown.

MakeoverMonday: Facebook’s Energy Footprint

My makeover. Click here to download the workbook (requires Tableau v10)

A first for MakeoverMonday: I ended up pretty much remaking the original chart, with only small tweaks. Once I’d locked onto the story I wanted to tell, I couldn’t escape the fact that Facebook’s original version of this chart was pretty much just right. In fact, it’s possible that I’ve complicated the message with my version.

How did I get to my version?

The original.
The original.

I thought Facebook’s whole report was fascinating. I learnt a lot from this graphical report.

As I explored the data, and cross-reference the report, it all seemed to hone in on the amount of renewable energy being used. The power usage itself is interesting, but the ambition for Facebook is to get the CaRE up to 50% by 2018 (and ultimately 100%).

Too sparse
Too sparse

I tried to draw a slope chart first, but it looked too sparse. Also, it hid what I thought was some important information – the volatility of CaRE:

Clean and renewable is highly volatile
Clean and renewable is highly volatile

I wanted to pursue this volatilty because it’s hard to say there’s a long-term trend upwards for CaRE when there was such a big trough in 2013. Unfortunately, I couldn’t find that information in the report.

Without the information on the volatility, I figured I’d accept Facebook’s word and focus on Facebook’s hitting the 25% CaRE by 2015 goal. As I drew different versions, it seemed that only a line chart or an area chart with CaRE along the baseline made the point. The other energy types are secondary information: as long as CaRE is going up, I don’t really care too much what’s happening to coal and nuclear.

What do I like about the original?
  1. Annotations on the marks explain the data
  2. Headline on the left summarises the point being made
  3. Forecast line has a different format
What don’t I like about the original?
  1. The x-axis year labels aren’t horizontal, and they don’t align very well to the marks themselves
  2. The y-axis % scale only goes up to 50%. On the one hand, this is fine, because it fits the range of the data. On the other hand, 35% means that 65% of data is still not renewable.
My changes


  • I used an area chart, with faded colours for all but CaRE to add context to the main story about CaRE usage
  • This choice also forced the y-axis to go from 0-100%. Now you can see that while goals are being hit, companies with huge data centres still have a long way to go.
  • I added a reference line for the 2015. This helps imply that the goal is continuous. The goal doesn’t stop in 2015.